Automatic Feedback Models to Students Freely Written Comments

Abstract

Teachers and professors, in different educational institutions, always wanted to grasp the learning experience of their students so they can give them proper guidance and intervene when it is necessary. However, tracking every student can be very challenging. In this context, we created an online questionnaire and asked students to fill it after each lesson. The Professor read what the students had written to get a better understanding of their learning experience and to reply to them with the adequate guidance. However, it turns out that professors quickly become overwhelmed and students have to wait longer before receiving any feedback. In this paper, we describe our method of building an automatic feedback model that will be used to help professors. We tried two approaches of building the models with or without a padding of the context. Empirical results show that our padded models can achieve 0.664 micro F-score